J. Am. Chem. SOC.1994,116, 10914-10920
10914
Effect of Phosphate Activating Group on Oligonucleotide Formation on Montmorillonite: The Regioselective Formation of 3’,5’-Linked Oligoadenylates K. Joseph Prabahar, Timothy D. Cole, and James P. Ferris* Contribution ffom the Department of Chemistry, Rensselaer Polytechnic Institute, Troy, New York 12180-3590 Received June 27, 1994@
Abstract: The effects of amine structure on the montmorillonite-catalyzedoligomerization of the 5’-phosphoramidates of adenosine are investigated. 4-Aminopyridine derivatives yielded oligoadenylates as long as dodecamers with a regioselectivity for 3’,5’-phosphodiester bond formation averaging 88%. Linear and cyclic oligomers are obtained and no A5’ppA-containing products are detected. Oligomers as long as the hexanucleotide are obtained using 2-aminobenzimidazole as the activating group. A predominance of pATpA is detected in the dimer fraction along with cyclic 3’,5’-trimer; no AsppA-containing oligomers were detected. Little or no oligomer formation was observed when morpholine, piperidine, pyrazole, 1,2,4-triazole, and 2-pyridone are used as phosphate-activating groups. The effects of the structure of the phosphate activating group on the oligomer structure and chain lengths are discussed.
triphosphates of adenosine do not condense to form oligomers on montmorillonite.’ More recently, while this research was in progress, Sawai and co-workers reported the use of 2methylbenzimidazole, triazole, and 3-nitrotriazole as 5‘-phosphate activating groups for the UOZ*+-catalyzed formation of 2’3’-linked oligonucleotides. l 3 The oligomers formed using these leaving groups were shorter than those formed using imidazole. In the present study we report the effect of phosphorus leaving group on oligomer formation and on the regioselectivity of the phosphodiester bond formation in the presence of montmorillonite.
Introduction
The observation that RNA can catalyze reactions as well as store genetic information suggests that it may have been the most important biopolymer in the earliest life on earth.’,* This scenario implies that the organic precursors to RNA-like molecules formed spontaneously on the primitive earth and condensed to form oligonucleotides. The information in these oligomers was preserved by template-directed synthesis of the complementary RNA oligomers. Problems with this scenario include the failure to find plausible prebiotic synthesis of mononucleotides and polynucleotides and the limited capability of template-directed s y n t h e s i ~ . ~Recently, .~ studies from our laboratory emphasized the central role of catalysis in prebiotic Experimental Section chemistry5 and have demonstrated the formation of oligoadenylates by the condensation of 5‘-phosphorimidazolide of General. lH NMR spectra were recorded on Varian XL-200 and Varian Unity 500 instruments. 13CNMR and 31PNMR were recorded adenosine (ImpA) on montmorillonite in aqueous s ~ l u t i o n . ~ ~ ~ on Varian Unity 500 instrument operating at 125 and 202 MHz, The condensation of the dinucleotide pyrophosphate of the respectively. NMR spectra were obtained as D20 solutions using the adenosine (A5‘ppA) with ImpA and ImpU results in the following references; 3-(trimethylsilyl)-l-propanesulfonic acid (Tsp) formation of oligomers in which over 80% of the phosphodiester for ’H NMR, DMSO-& solvent peaks for I3C NMR, and 85% H3P04 bonds are 3’,5‘-linked.5~8~9 Oligomers synthesized by these and for 31PNMR spectra. The chemical shifts are reported in ppm. Highother catalytic processes may have been replicated and elongated resolution mass spectra (FAB; matrix of dithiothreitol and dithioerythby template directed processes.lo,ll ritol) were obtained at the School of Chemical Sciences, University of The role of the phosphate activating group in oligonucleotide Illinois, Urbana. The CISBondapak reverse-phase gel 100 A (mesh formation on RNA templates or on montmorillonitehas not been 15-20 pm) was purchased from Waters for use in the preparative investigated in a systematic fashion. Studies on the nonenzyreverse-phase column. The Dowex 50 W-X8 cation exchange resin (mesh 15-20 pm) was purchased from Bio-rad laboratories and matic template directed synthesis of oligonucleotides utilize activated by Cooper’s pr0~edure.l~ Adenosine 5’-monophosphate (5‘imidazole or substituted imidazoles as leaving groups because AMP), l-ethyl-3-(3-(dimethylamino)propyl)carbodiimide(EDAC), and nucleoside di- and triphosphates react slowly and undergo ribonuclease Tz (RNase T2) were obtained from Sigma. Bacterial hydrolysis.’* We have shown that AS’ppA and the di- and Abstract published in Advance ACS Abstracts, November 1, 1994. (1) Been, M. D.; Cech, T. R. Science 1988,239, 1412. (2) Guemer-Takoda, C.; Gardiner, K.; Marsh, T.; Pace, N.; Altman, S. Cell 1983,35, 849. (3) Ferris, J. P. Cold Spring Harbor Symp. Quant. Biol. 1987,52, 29. (4) Joyce, G. F.; Inoue, T.; Orgel, L. E. J. Mol. Biol. 1984,176, 279. (5) Ferris, J. P.; Ertem, G. Origins Life Evol. Biosphere 1993,23, 229. (6) Ferris, J. P.; Ertem, G. Science 1992,257, 1387. (7) Fems, J. P.; Ertem, G. J. Am. Chem. SOC. 1993,115, 12270. (8) Ferris, J. P.; Ertem, G. Origins Life Evol. Biosphere 1992,22, 369. (9) Ding, Z. P. Ph.D. Thesis, Rensselaer Polytechnic Institute, Troy, NY, 1993. (10) Li, T.; Nicolaou, K. C. Nature 1994,369, 218. (11) Severs, D.; Von Kiedrowski, G. Nature 1994,369, 221 @
alkaline phosphatase (APH) was obtained from Sigma and Worthington Biochemical Corp. 2,2’-Dipyridyl disulfide, triphenylphosphine, 4-(dimethylamino)pyridine, 4-aminopyridine, 2-aminobenzimidazole, morpholine, piperidine, and 2-pyridone were obtained from Aldrich. DMF and DMSO were purchased from Fisher, and ether, acetone, and triethylamine were obtained from Mallinckrodt. 4-(Methylamino)pyridine was synthesized and had a mp of 123-124 “C (lit. mp 124.5-
(12) Wiemann, B. J.; Lohrmann, R.; Orgel, L. E.; Schneider-Bemloehr, H.; Sulston, J. E. Science 1968,161, 387. (13) Sawai, H.; Shibusawa, T.; Kuroda, K. Bull. Chem. SOC. Jpn. 1990, 63, 1776. (14) Cooper, T. G. The Tools of Biochemistry; John Wiley and Sons: New York, 1977.
0002-7863/94/1516-10914$04.50/00 1994 American Chemical Society
Formation of 3',5'-Linked Oligoadenylates 125 OC).I5 Montmorillonite 22A clay was obtained from Wards Natural Science Establishment,I6and the homoionic montmorillonite 22A was prepared by the saturation method.'? HPLC analyses were performed on a Waters HPLC system equipped with Lamda-Max model 481 W detector operating at 260 nm, on a p-Bondapak CISreverse-phase column using a gradient of 0.005 M NaHzP04 in 5% methanol at pH 3.5 mixed with 0.01 M NaHzP04 in 40% methanol at pH 4.0 and on a HEMA-IEC BIO Q anion-exchange column from Alltech using a gradient of 0-0.4 M NaC104 at pH 8 with 2 mM Tris buffer. No Tris was used when samples were collected for further analysis and the column was eluted at isocratic mode of buffer A 98% and buffer B 2% for 6 min followed by a regular gradient elution mode. General Procedure for the Preparation of Activated Nucleotides 3a-f.I8 A mixture of 5 ' - m P H z O (free acid) (0.365 g, 1 mmol) and heterocyclic base 2a-f (3 mmol) was dissolved in DMF (10 mL) and DMSO (5 mL) in a 50 mL flask, and the solvents were evaporated to 2 mL at a reduced pressure to remove H20. The evaporation was repeated twice with DMF (2 x 10 mL). The residue was dissolved in a mixture of DMF (10 mL) and DMSO (10 mL) and stirred with 2,2'dipyridyl disulfide (0.666 g, 3 mmol), triphenylphosphine (0.786 g, 3 mmol), and triethylamine (1 mL) under an argon atmosphere for 2 h. The resulting clear yellow reaction mixture was added dropwise to a 1 L flask containing a solution of anhydrous sodium perchlorate (2 g) in a mixture of ether (200 mL), acetone (125 mL), and triethylamine (15 mL) with stirring under argon atmosphere. The stirring was continued for 2 h, and a colorless, flocculant solid separated. The stirring was stopped and the solid allowed to settle for 15 min. The supematent was drained and the remaining portion centrifuged. The resulting colorless pellet was washed twice with 50% acetone-ether mixture, centrifuged, and dried under vacuum. General Procedure for the Preparation of Activated Nucleotides (3g-h).I9 A mixture of 5'-AMPH20 (free acid) (0.365 g, 1 mmol), heterocyclic base 2g-h (3 mmol), and l-ethyl-3-(3-(dimethylamino)propy1)carbodiimide hydrochloride (EDAC) (0.955 g, 5 mmol) was dissolved in water (3 mL). The solution was adjusted to pH 5-5.5 with sodium hydroxide solution and stirred at room temperature for 4 h. During the reaction, the EDAC was hydrolyzed to a substituted urea. The activated nucleotide was separated from the urea derivative by passing the reaction mixture through a Dowex 50 W-X8 cationexchange column. The column was eluted with water (150 mL), and the collected fraction was lyophilized to yield a colorless solid. General Procedure for the Purification of Activated Nucleotides (3). The activated nucleotides 3a-h were purified using preparative p-Bondapak reverse-phase column. The column was eluted with water and water-acetonitrile mixture. The pH of the eluents was adjusted to 8-9 with a trace amount of triethylamine; 10 pL of triethylamine is sufficient for 1000 mL of water. Excess triethylamine leads to the formation of triethylammonium salts of the activated nucleotides. The chromatography was performed at 4 "C and the column was eluted under 5-10 psi argon pressure with the flow rate of 7-10 mL/min. Fractions (100 mL) were collected and analyzed by reverse-phase HPLC. The fractions which contained the activated nucleotides were pooled and lyophilized to yield a colorless solid. The activated nucleotides were obtained as sodium salts. Adenosine 5'-Phosphoro-4-(dimethylamino)pyridinium (4-(CH3)2NpypA) (3a). Compound 3a was shown to be 95% pure by HPLC on a p-Bondapak column. It was obtained as a colorless solid (0.35 g, 77.6%). 'H NMR (D20): 6 3.05 (s, 6H), 4.35 (m, 2H), 4.52 (m, lH), 4.7 (t, lH), 4.8 (merged with HOD peak), 5.9 (d, lH), 6.4 (d, 2H) 7.95 (m, 2H), 8.15 (s, lH), 8.17 (s, 1H). I3C NMR (DzO): 6 157.9, 156.5, 154.1, 150.0, 141.9, 119.8, 107.5, 89.2, 84.4, 74.3, 71.1, 68.5, 40.9. 31PNMR (D20): 6 -5.21. Compound 3a was not sufficiently stable (15) Wibaut, J. P.; Broekman, F. W. Red. Trav. Chim. Pays-Bas 1961, 80, 309.
(16) American Petroleum Institute: Clay Mineral Standards, American Petroleum Institute, Project 49, Preliminary Report 7B, Chemical Analysis; Columbia University: New York, 1951. (17) Brindley, G. W.; Ertem, G. Clays Clay Miner. 1971,19, 399. (18) Mukaiyama, T.; Hashimoto, M. Bull. Chem. SOC. Jpn. 1971, 44, 2284. (19) Ivanovskaya, M. G.; Gottikh, M. B.; Sabarova, Z. A. Nucleosides Nucleotides 1987,6, 913.
J. Am. Chem. SOC., Vol. 116, No. 24, 1994 10915 at room temperature to submit for FAB MS analysis. It was stable for 3 weeks at -20 "C. Adenosine 5'-Phosphoro-4-(methylamino)pyridinium (4-CH3NHpypA) (3b). Compound 3b was shown to be 97% pure by HPLC on ap-Bondapak column. It was obtained as a colorless solid (0.33 g, 76.5%). 'H NMR (D2O): 6 2.8 (s, 3H), 4.3 (m, 2H) 4.55 (m, lH), 4.85 (merged with HOD peak), 5.9 (d, lH), 6.15 (m, 2H), 7.8 (m, lH), 7.9 (m, lH), 8.05 (s, lH), 8.10 (s, 1H). I3C NMR (DzO): 6 159.6, 156.7, 154.3, 150.2, 143.8, 141.87, 141.22, 119.9, 110.6, 105.2, 89.1, 84.4, 74.4, 71.2, 68.4, 30.4. 31PNMR (D20): 6 -1.07. Compound 3b was not sufficiently stable at room temperature to submit for FAB MS analysis. It was stable for 3 weeks at -20 "C. Adenosine 5'-Phosphoro-4-aminopyridinium(4-NH2pypA)(3c). Compound 3c was shown to be 76% pure by HPLC on a p-Bondapak column. It was obtained as a colorless solid (0.30 g, 70.9%). 'H NMR (DzO): 6 4.25 (m, 2H), 4.5 (m, lH), 4.9 (merged with HOD peak), 5.85 (d, lH), 6.35 (d, 2H), 7.9 (m, 2H), 8.15 (s, lH), 8.2 (s, 1H). 31P NMR (DzO): 6 -1.07. Compound 3c was not sufficiently stable to submit for FAB MS analysis. It was stable for 2 weeks at -20 "C. Adenosine 5'-Phosphoro-2-aminobenzimidazolide (2-NHzbenzimpA) (3d). Compound 3d was shown to be 98% pure by HPLC on a p-Bondapak column. It was obtained as a colorless solid (0.375 g, 81.1%). 'H NMR (DzO): 6 4.05 (m, 2H), 4.25 (m, lH), 4.28 (t, lH), 4.65 (t, lH), 5.9 (d, lH), 6.7 (t, lH), 6.9 (t, lH), 7.0 (d, lH), 7.2 (d, lH), 7.99 (s, lH), 8.0 (s, 1H). 13C NMR (DzO): 6 157.4, 157.3, 154.0, 150.0, 141.1, 134.0, 124.6, 122.3, 120.4, 115.3, 114.2, 88.2, 84.8, 75.1, 71.7, 67.0. 31PNMR (DzO): 6 -7.36. High-resolution MS (positive-ion FAB): calcd for CI~HZONSO~P [as acid form, M HI 463.1243, found 463.1250. Adenosine (5'-Phosphoro-4-morpholinide)(MorpA) (3e).I8 Compound 3e was shown to be 99% pure by HPLC on a p-Bondapak column. It was obtained as a colorless solid (0.36 g, 86.7%). IH NMR (DzO): 6 2.9 (t, 2H), 3.55 (t, 2H), 4.05 (m, 2H), 4.35 (m, lH), 4.5 (t, lH), 4.8 (merged with HOD peak), 6.1 (d, lH), 8.25 (s, lH), 8.45 (s, 1H). I3C NMR (Dz0): 6 156.9, 154.6, 150.47, 141.2, 119.9, 88.6, 85.5, 75.7, 71.9, 68.4, 65.3, 46.3. 31PNMR (Dz0): 6 8.05. Highresolution MS (positive-ion FAB): calcd for C14HzlN607PNa [MH] 439.1107, found 439.1111. Adenosine (5'-Phosphoro-1-piperidinide)(PiperpA) (30. Compound 3f was shown to be 98% pure by HPLC on a p-Bondapak column. It was obtained as a colorless solid (0.31 g, 74.8%). IH NMR (DzO): 6 1.4 (bs, 6H), 2.85 (bs, 4H), 4.05 (m, 2H), 4.37 (m, lH), 4.55 (t, lH), 4.8 (merged with HOD peak), 6.1 (d, lH), 8.22 (s, lH), 8.45 (s, 1H). I3C NMR: (Dz0) 6 159.7, 154.2, 150.3, 141.2, 119.8, 88.8, 85.5, 75.9, 71.9, 65.3, 27.4, 25.6. 31PNMR (DzO): 6 9.69. Highresolution MS (positive-ion FAB): calcd for C15H23N606PNa [MH] 437.1314, found 437.1323. Adenosine 5'-Phosphoropyrazolide(PyrapA) (3g). Compound 3g was shown to be 96% pure by HPLC on a p-Bondapak column. It was obtained as a colorless solid (0.25 g, 62.9%). 'H NMR (D20): 6 4.12 (m, 2H), 4.3 (m, lH), 4.38 (t, lH), 4.7 (t, lH), 6.0 (d, lH), 6.35 (bs, lH), 7.71 (bs, lH), 7.91 (bs, lH), 8.11 (s, lH), 8.19 (s, 1H). I3C NMR(Dz0): 6 157.0, 154.3, 150.5, 145.6, 141.1, 136.3, 120.1, 108.6, 88.6, 85.0, 75.7, 72.0, 67.2. "P NMR (DzO): 6 -6.91. Highresolution MS (positive-ion FAB): calcd for C13H1&06PNa [MH] 420.0797, found 420.0810. Adenosine 5'-Phosphoro-1,2,4-triazolide(TriazpA) (3h). Compound 3h was shown to be 98% pure by HPLC on a p-Bondapak column. It was obtained as a colorless solid (0.26 g, 62.5%). IH NMR (D20): 6 4.2 (m. 2H), 4.3 (m, lH), 4.45 (t, lH), 4.75 (t, lH), 6.05 (d, lH), 8.0 (s, lH), 8.22 (s, lH), 8.3 (s, lH), 8.6 (s, 1H). I3C NMR (Dz0): 6 157.2, 154.7, 150.6, 150.0, 141.4, 120.2, 88.2, 84.8, 75.4, 71.8,67.7. 31PNMR (D20): 6 -9.33. High-resolution MS (positiveion FAB): calcd for C I Z H I ~ N S ~[MH] ~ P N421.0750, ~ found 421.0751. Adenosine (5'-Phosphoroxy-2-pyridinide)(2-OxypyripA) (3i). Compound 3i was shown to be 99% pure by HPLC on a p-Bondapak column. It was obtained as a colorless solid (0.25 g, 56%). IH NMR (DzO): 6 4.25 (m, 2H), 4.38 (m, lH), 4.5 (t, IH), 4.85 (lH, merged with HOD peak), 6.05 (d, lH), 7.0 (m, 2H), 7.55 (m, IH), 8.05 (m, lH), 8.10 (s, lH), 8.25 (s, 1H). 31P NMR (DzO): 6 -5.32. Highresolution MS (positive-ion FAB): calcd for C I ~ H I ~ N ~ O[MH] ~PN~ 447.0794, found 447.0806.
+
10916 J. Am. Chem. SOC.,Vol. 116, No. 24, 1994
Prabahar et al.
1
1.55e+l
1 0
20
40
60 Time (h)
80
100
Figure 2. Pseudo-first-order plot of the kinetic data from the hydrolysis 3a in 0.2 M NaCl, 0.075 M MgC12, and 0.1 M HEPES at pH 8.1 at 23-25 "C.
Figure 1. Anion-exchange HPLC of the reaction products of 3a on Na+ montmorillonite 22A. General Procedure for the Clay-Mediated Oligomerization Reaction of Activated Nucleotides. An electrolyte solution (50 mL) was prepared from a mixture of sodium chloride (0.2 M) and magnesium chloride (0.075 M), and the pH was adjusted to 8. A stock solution (2.5 mL) of activated nucleotide (0.015 M) was prepared from the sodium chloride-magnesium chloride electrolyte. The activated nucleotide solution (1 mL) was added to Na+ montmorillonite 22A clay (0.05 g); the reaction mixture was vortexed for 30 s and the pH adjusted to 8. The reaction mixture was kept at the room temperature for 7 days. The activated nucleotide solution (1 mL) which did not contain montmorillonite was kept under the same conditions as a control reaction. The reaction with the clay was centrifuged, and the supematant was removed from the clay and filtered through a 0.45 pm pore filter. The montmorillonite clay was vortexed with ammonium acetate (1 mL, 0.1 M) solution and allowed to stand for 24 h. The reaction mixture was centrifuged. The supematant was filtered and combined with the f i s t supematant solution. Analysis of Oligomers Formed from the Reaction of 4-(CH& NpypA (3a) on Na+ Montmorillonite. The product mixture from the reaction of 3a (45 pL) were analyzed by anion-exchange HPLC (Figure 1); the monomer, dimer, trimer and tetramer fractions were collected and stored in the freezer. An aliquot of each fractions was reinjected in the anion-exchange column to determine the homogeneity of the collected fractions. The collected monomer, dimer, and trimer fractions were analyzed by reverse-phase HPLC. The PA, pA2pA, pA3'pA, pA3'pA3'pA, and pAypA2'pA were identified from their retention times in the reverse-phase HPLC profile and the presence of these isomers further confirmed by co-injecting the authentic samples along with the collected fraction in the reverse-phase column by HPLC. The ratio of oligomers containing terminal phosphate were determined by APH. An aliquot amount of each fraction (300 pL) was treated with 0.166 units of APH at 37 "C for 4 h. The monomer and dimer hydrolysis products were analyzed by reverse-phase HPLC, and trimer and tetramer hydrolysis products were analyzed by anionexchange chromatography. The ratios of 3',5'- and 2',5'-linkages in the monomer through tetramer fractions were determined by ion-exchange HPLC analysis of the RNase TZand RNase Tz followed by APH reaction products of the collected fractions. (i) The monomer fraction (300 pL) was treated with 0.3 units of RNase TZat 37 "C for 4 h and analyzed by reverse-phase HPLC. (ii) The dimer fraction (300 pL) was treated with 0.3 units of RNase Tz, incubated at 37 "C for 3 h, and analyzed on the reverse-phase and anion-exchange HPLC. (iii) The dimer fraction RNase TZ reaction mixture (300 pL) was adjusted the pH to 8, treated with 0.166 units of APH at 37 "C, and analyzed by reverse-phase and anion-exchange HPLC. (iv) The trimer fraction (300 pL) was treated with 0.8 units of RNase Tz at 37 "C for 4 h and analyzed by anion exchange and reverse-phase HPLC.
(v) The trimer RNase TZreaction mixture (300 mL) was adjusted to pH 8, treated with 0.166 units of APH at 37 "C for 2 h, and analyzed using anion-exchange and reverse-phase chromatography. (vi) The tetramer fraction (300 pL) was treated with 0.3 units of RNase Tz at 37 OC for 2 h and analyzed in the anion-exchange chromatography. (vii) The tetramer fraction RNase TZreaction mixture (300 pL) was adjusted the pH to 8, treated with 0.166 units of APH at 37 "C for 2 h and analyzed by anion-exchange and reverse-phase chromatography. Isolation of Cyclic Trimer from the Reaction Products of 4-(CH&NpypA (3a) by HPLC. The reaction products of 4-(CH3)2NpypA 3a (45 pL) were injected on the anion-exchange column and eluted the column at isocratic mode of buffer A 98% and buffer B 2% for 6 min followed by a regular gradient elution mode. The dimer fraction was resolved into three peaks, the first peak was collected as pAZ'pA(fraction A), and the remaining two peaks of the dimer fraction were eluted close together and collected together as a mixture of cyclic trimer and pA3'pA (fraction B). The presence of pAZ'pAin the fraction A and cyclic trimer and pA3'pA in fraction B was confirmed by analyzing the collected fractions by reverse-phase HPLC. The cyclic trimer was characterized by HPLC analysis of the RNase Tz and the RNase Tz plus APH hydrolysis products. Isolation and Characterization of Dimer Fraction from the Reaction Products of 2-NHzbenzimpA(3d)on Na+ Montmorillonite. The reaction products of 3d (45 pL) were separated by the anionexchange HPLC, and the dimer fraction was collected. The dimer fraction was reinjected in the anion-exchange column to determine its homogeneity and then was injected on the reverse-phase column, and the pAZ'pA, cyclic trimer, and pA3'pA were identified by their HPLC retention times. The products were characterized by HPLC analysis of the RNase TZ and the RNase TZplus APH hydrolysis products. Hydrolysis of 4-(CH&NpypA (3a) at pH 8. A buffer solution of 0.2 M NaCI, 0.075 M MgC12, and 0.1 M of (N(2-hydroxyethy1)piperazine-A'-(2-ethanesulfonicacid) (HEPES) was prepared, and the pH was adjusted to 8; 1 mL of 4-(CH&NpypA 3a solution (0.015 M) was prepared using HEPES buffer and the hydrolysis reaction was carried out at 23-25 "C. The hydrolysis of 3a was monitored at regular intervals by analyzing the reaction mixture (10 pL) by reverse-phase HPLC. A linear plot of In [4-(CH3)~NpypA]vs time (Figure 2) was obtained. General Procedure for the Binding Studies of 3a and ImpA on Na+ Montmorillonite. A solution of activated nucleotide (2 mL, 0.015 M) was prepared in 0.2 M NaCl, 0.075 M MgC12, and 0.1 M HEPES (pH 8) buffer at 4 "C; 1 mL of this solution was added to Na+ montmorillonite (50 mg) and vortexed for 30 s. The reaction mixture was centrifuged at regular intervals, and 5 p L was analyzed by anionexchange HPLC. A plot of HPLC peak area of the activated nucleotide vs time is given in Figures 3 and 4. Molecular Modeling Study of 4-(CH3)2NpypApApA. The structure of 4-(CH3)2NpypApApA was modeled in the CAChe molecular modeling work system (release 3.5) running on a Macintosh Centris 650, and the molecular mechanics calculation was carried out using MM2 force field parameters. Conjugate gradient was used to locate the energy minimum, and all atoms were moved at once. Van der Waals interactions between atoms separated by greater than 9.00 A
J. Am. Chem. SOC., Vol. 116, No. 24, 1994 10917
Formation of 3’,5’-Linked Oligoadenylates
Scheme 1
HPLC peak
HO OH
HO OH 3a. R.R = CH, Jb. R= H, R = CH3 3c. R.R = H
1
y 2
Time ( m i d
Figure 3. Loss of 3a (0.015 M) in the presence of Na+ montmorillonite 22A at pH 8 and 4 “C measured by HPLC peak area at 260 nm (A). Total oligomer formation not corrected for hyperchromicity measured at 260 nm (A). The intensity of the of the UV absorption of 3a is
ri
EDAC
HO OH
” I I
HO OH
1
3 X or XH 2 -
greater than that of the oligomers because of the contribution of the DMAP activating group to the absorption of 3a.
4.00e+7
3.ooe+7
*3j was purchased from Sigma.
HPLC peak area
using preparative p-Bondapak reverse-phase chromatography. To minimize hydrolysis to pA during purification, chromatography was performed at 4 “C under 5-10 psi argon pressure with the flow rate of 7-10 mL/min. The water used as eluent was adjusted to pH 9 with a trace of triethylamine to minimize O.OOe+O 1 0 10 20 30 40 50 60 70 80 90 100 the hydrolysis of the activated nucleotides. Use of an excess Time ( m i d of triethylamine was avoided since this resulted in the formation Figure 4. Loss of ImpA (0.015 M) in the presence of Na+ montmoof triethylammonium salts of the activated nucleotides. rillonite 22A at pH 8 and 4 “C measured by HPLC peak area at 260 Compounds 3a-i were characterized by ’H NMR, 13CNMR, nm (A).Total oligomer formation not corrected for hyperchromicity and 31P NMR spectroscopy. Compounds 3d-i were also measured at 260 nm (A). characterized by FAI3 high-resolution mass spectrometry. Mass spectra of compounds 3a-c were not obtained because they were excluded. The optimization was continued until the energy change are not stable at room temperature and are only stable for 2-3 was less than 0.001 kcal/mol. weeks at -20 “C. Results and Discussion Oligomerization Reaction of Activated Nucleotides 3a-j on Na+ Montmorillonite. The oligomerization reactions of Synthesis and Purification of Activated Nucleotides 3athe activated nucleotides 3a-j were carried out in NaC1-MgC12 f. Activated nucleotides 3a-f (Scheme 1) were synthesized electrolyte solution at pH 8 on montmorillonite 22A clay. The by the procedure of Mukaiyama et a1.8,18 The products were reactions of the activated nucleotides 3a-j in the absence of separated from the reaction mixture by precipitating as sodium montmorillonite were carried out as controls. The course of salts using a solution of sodium perchlorate in acetone-ether. each reaction was monitored by reverse-phase HPLC over a Although the synthesis of 3h was reported using the procedure period of 7 days. The analysis of reaction of compounds 3a-c of Mukaiyama et al., we were not able to prepare it or 3g and in the absence of montmorillonite by reverse-phase and anion3i by this route.13 The latter compounds were synthesized in exchange HPLC showed hydrolysis was the main reaction aqueous solution using l-ethyl-3-(3-(dimethylamino)propyl)pathway. Diadenosine pyrophosphate (AsppA) (6%) and dimer carbodiimide hydrochloride (EDAC) as condensing agent.l9 The and trimer isomers were also detected as products. Compound optimal reaction time was determined by monitoring the reaction 3d was very stable in the absence of montmorillonite; only 7% using reverse-phase HPLC. The activated nucleotide was hydrolyzed to pA over 7 days. Compounds 3e-j hydrolyzed separated from the urea derivative formed by the hydrolysis of to pA in the absence of montmorillonite and no oligomers or EDAC by passing the reaction mixture through a Na+-Dowex A5’ppA were detected as hydrolysis products. 50 cation-exchange column and washing with water. The water Products from the reaction of compounds 3a-c on montmowas lyophilized to yield the activated nucleotide as sodium salt. rillonite were analyzed by anion-exchange and reverse-phase It was not possible to prepare N-phosphoryl-5’-derivativesof HPLC. Oligomers containing up to 12 monomer units were 2-aminopyridine, 2-(dimethy1amino)pyridine and 1,1,3,3-tetdetected by anion-exchange HPLC (Figure 1).21 Oligomers ramethylguanidine by either of the above procedures. Steric containing up to six monomer units were obtained by the effects of the 2-substituent may explain our failure to isolate reaction of compound 3d in the presence of montmorillonite N-phosphoryl-5’-derivativesof 2-amino- and 2-(dimethylamino)Hydrolysis to PA was the main reaction pathway of 3e-j on pyridine.20 Compounds 3a-h were shown to be 70-98% pure by (20) Hassner, A.; Krepski, L. R.; Alexanian, V. Tetrahedron 1978,34, reverse-phase HPLC, and further purification was carried out 2069. 2.00e+7
10918 J. Am. Chem. SOC., Vol. 116, No. 24, 1994
Prabahar et al.
Table 1. Products from the Reaction of 3e-j in the Presence of Montmorillonite“
2-Aminobenzimidazole was studied as a leaving group because it contains the imidazole moiety which is known to yield oligomers in the presence of montmorillonite. It was activated nucleotides pA AS’ppA pA2’pA pA3’pA postulated that it may be more effective than imidazole because MOTA (3e) 58 1.8 1.o 0 the 2-amino grouping enhances the basicity of the imidazole 65 0 1.5 0.8 PiperpA (30 by generating a substituted guanidine derivative (pKa7.5).28The 3.9 1.8 PyrazpA (3g) 88 2.6 TriazpA (3h) 96 4.0 0 0 corresponding phosphoramidate would be protonated more 2-OxypyripA (3i) 75 1.o 0.1 0.8 extensively by the acidic montmorillonite and either undergo facile oligomer formation or hydrolysis. Oligomers which ” Determined from the products of HPLC analysis on reverse-phase after APH hydrolysis. contain almost comparable amounts of 2’,5’- and 3’,5’-phosphodiester bonds are formed using 3d (see below), but in yields montmorillonite. Low yields of A5’ppA and dinucleotides were lower than those observed with imidazole itself. These low also detected (Table 1). yields may reflect the different steric requirements of 3d as The Effect of Phosphate Activating Group on Nucleotide compared to ImpA. For example, small changes in the Reactivity. The nucleophilic displacement of amines from substituents present on the imidazole ring can result in major phosphoramidates proceeds in a concerted p r o ~ e s s . ~The ~ , ~ ~ changes in the structures of the oligo(G)s formed by template position of the transition state depends on the relative nucleodirected synthesis on p01y(C).~~ Alternatively, the low yields philicities of the nucleophile and leaving group. The rate of may reflect the inefficient formation of 2’,5’-linked phosphodihydrolysis of the monoanions of phosphoramidates decrease in ester bond from adenine nucleotides on montmorillonite. For the order shown for the following amines: pyridine > aniline example, the presence of 2’,5’-linked unit on the end of the > ammonia =- imidazole > n-butylamine > 4-(dimethylamino)growing oligomer chain inhibits chain elongation in the oligopyridine (DMAP).24 Linear correlations of the log of the rate merization of 1 m ~ A Similarly, .~ binding U022f to montmorilconstant with amine pKa are observed only within groups of lonite inhibits the formation of longer 2’,5’-linked oligomers; structurally related amines. The substantially greater stability longer oligomers were formed when the same amount of U0z2+ of imidazole and DMAP phosphoramidates as compared to the was used as a catalyst in the absence of montmorillonite.8 corresponding pyridine derivative reflects the greater delocalThe 4-aminopyridine activated derivatives of 5’-AMP (3aization of the positive charge on the amino grouping in the c) are the most effective compounds for the oligo(A) formation former compounds.25 The charge is delocalized to the second on montmorillonite. Chain lengths as high as the dodecamer nitrogen by resonance interaction and by solvation.26 (Figure 1) and 3‘5‘-phosphodiester bond regioselectivity of A variety of amine structural types were investigated in the about 88% were observed (see below) using the (dimethylamisearch for new phosphate activating groups because of the no)pyridine (DAMP) derivative 3a. The successful synthesis absence of a general correlation between hydrolytic reactivity of oligomers with ImpA6s7 and 3a may reflect the greater and amine basicity. The reaction of the phosphorimidazolide stability of the activated monomers in aqueous solution, and of adenosine is believed to be initiated by protonation of the their enhanced reactivity on the montmorillonite surface due to imidazole (pKa = 6.0)27grouping.* Since the pKa of phosphothe lower concentration of water and the higher concentration ramidate is approximately 1 pKa unit less than the pKa of the of monomer. Acid catalysis by montmorillonite may also corresponding amine, (e.g., the pKa of imidazole is 7.0)28it was contribute to oligomer formation. Compound 3a was used in expected that phosphoramidates of strongly basic amines such the studies on oligomer formation because DMAP is comas morpholine (3e) and piperidine (30 would be protonated more mercially available and the activated nucleotide is more stable readily than the phosphorimidazolide grouping and undergo than the 4-aminopyridine derivative 3c.31 rapid reaction.24 Facile hydrolysis of 3e and 3f but no oligomer Our success in forming oligonucleotides when imidazoles or formation was observed. The 5‘-phosphoramidate of adenosine 4-aminopyridines are the activating groups of 5’-AMP, and our (3j) is resistant to hydrolysis, a finding consistent with the low failure with other activating groups makes it possible to draw basicity of ammonia and the reported slow hydrolysis of the some conclusions concerning the optimal properties of activated phosphoramidate of ammonia.25 Pyrazole and triazole are weak nucleotides. First, nucleotides with positively charged leaving bases (pKas 2.5 and 2.3, respectively),28 so the corresponding groups bind more efficiently to the catalytic sites on the phosphoramidates 3g and 3h, respectively, are not basic and negatively charged faces of montmorill~nite.~~ Binding studies are not protonated by montmorillonite. They undergo rapid with 3a demonstrate its rapid and almost complete binding to hydrolysis because pyrazole and triazole are good leaving groups montmorillonite (Figure 3). ImpA, which must be protonated even in the absence of acid catalysts.29 The 2-oxypyridinyl by acidic sites on the montmorillonite, binds less strongly. activated nucleotide 3i is formally a phosphate ester, but its Second, leaving groups consisting of resonance stabilized displacement may be energetically favored because of 2-pyrications, such as those present on 3a-d and protonated ImpA done formation. The reaction proceeded, but only hydrolysis initiate oligomer formation. The absence of oligomer formation to pA was observed. when 3e and 3f react on montmorillonite indicates that when (21) Stribling,R. J. Chromatogr. 1991,338,474.Anion-exchange HPLC the positive charge is not delocalized oligomers are not formed separates compounds principally on the basis of the number of charges in from the activated monomer. Third, the ratio of the rate of the molecule. Dinucleotides of structure pApA and cyclic trinucleotides oligomer formation to the rate of hydrolysis of the activated both have three negative charges and consequently appear in the fraction designated as “dimers”. monomer is a better indicator of the extent of oligomer formation (22)Bome, N.; Williams, A. J. Am. Chem. SOC. 1984, 106, 7591. than simply the rate of monomer h y d r ~ l y s i s .Compound ~~ 3a (23) Skoog, M. T.; Jencks, W. P. J. Am. Chem. SOC.1984, 106, 7597. has a half-life of hydrolysis of 1 day (see below) while that of (24) Benkovic, S. J.; Sampson, E. J. J. Am. Chem. SOC.1971, 93, 4009. (25) Jencks, W. P.; Gilchrist, M. J. Am. Chem. SOC.1965, 87, 3199. (26) Hofle, G.; Steglich,W.; Vorbruggen, H. Angew. Chem., Int. Ed. Engl. 1978, 17, 569. (27) Kanavarioti, A.; Bemasconi, C . F.; Doodokyan, D. L.; Alberas, D. J. J. Am. Chem. SOC. 1989, 111, 7247. (28) Pemn, D. D. Dissociation Constants of Organic Bases in Aqueous Solutions; Butterworths Scientific Publications: London, 1965.
(29) Fife, T. H. Acc. Chem. Res. 1993, 26, 325. (30)Inoue. T.: Orpel, L. E. J. Am. Chem. SOC. 1981, 103, 7666. (31) Wakselman, 6.; Guibe’-Jampel,E. Tetrahedron Lett. 1970, 4715. (32) Fems, J. P.; Ertem, G.; Aganval, V. K. Origins Life Evol. Biosphere 1989, 19, 153. (33) Kanavarioti, A. Origins Life Evol. Biosphere 1986, 17, 85.
J. Am. Chem. Soc., Vol. 116, No. 24, 1994 10919
Formation of 3’,5’-Linked Oligoadenylates Table 2. Differentiation of the Cyclic and Linear Oligomers by APH Hydrolysis oligomer linear chain cyclic chain fraction oligomers (%) length oligomers (%) length A. Products from 3a “dimers” 37 2 63 3 “trimer” 86 3 14 4 “tetramers” 38 4 62 5 “pentamers” 65 5 35 6
B. Products from 3d 59 2
“dimers”
41
11
AP pA2’pA
C
73 8
89 11 C
9.5 35 9.5
C
15 2 45
C
C
C
C
C
2
C C
C
C
18 13 7
C
4
~2’p~p
c
C
PAP
8
c
p ~ ~ ’ p ~c p
C
4
62 38
A AZ‘PA
AP PA~’~A PAP
RNase TZ
RNase T2
16 2“ 71 b 11
+ APH
97 3“ b b b
a The source of the AZ’pAis unknown. Not expected as a reaction product.
Table 5. Comparison of Oligomers formed from 4-(CH&NpypA (3a) and ImpA
“dimers” “trimers” “tetramers” RNase RNase RNase RNase RNase RNase product T2” T2+APHa Tzb T2+APH0 Tzb T2+APHU ATpA
products
3
Table 3. Hydrolysis Products of the Oligomers from 3a
A
Table 4. Hydrolysis Products of the Cyclic Trimer-pAypA Mixture from 3a (%)
76 24 C
C C
3’,5’-links (%)
2a
m
c
& I
94
“Dimers“
p ~ 2 ’ p ~
,-[pA3pA3pA AP~AJ’PA
14 23
33 52
63
13 2.6
a
77
“Trimers” pA3’pAZA
Determined by reverse-phase HPLC. Determined by anion- J * [ exchange HPLC. Not expected. Not detected. LpA3pA3pA2pA3Ll AsppAJ’pA3’pA A5ppA3’pAz’pA (Ap)mAS’ppA(pA)n (m+n=2)
64 23
47
8
a
5
a
m
c
65
72
28
ImpA when measured under comparable reaction conditions is 15 a 4 a 9.3 days,34yet 3a gives higher yields of longer oligomers than L a ImpA because its rate of oligomerization on montmorillonite is much faster than that of ImpA (Figures 3 and 4). Half the 94 b ‘Tetramers“ oligomers are formed from 3a in about 1 h, while essentially no oligomers are formed from ImpA in that time period (Figures pA3’pA3’A3’pA 34 9 pA3’pAJ’AZpA 4 22 3 and 4). a 37 (pA)4 isomers Effect of Monomer Activating Group on Oligonucleotide LpA3pA3pA3pA3pA3j a 11 Structure. The oligomers formed by the reaction of 3a and L ~ A ~ P A ~ ~ A ~ P A ~43 ~ A ~ J 3d on montmorillonite were characterized by selective enzymatic a a 1 hydrolyses. The dimer through pentamer fractions from the A5’ppA(pA)3 a 32 (Ap)mA5’ppA(pA)n reaction of 3a were collected after separation by ion-exchange (mtn=3) HPLC. Each fraction was treated with APH, and the reaction mixture was analyzed by HPLC to determine the presence of Oligomer not detected. Not determined. Reference 7. isomers with terminal phosphate groupings (Table 2).5 The isomers which contained 3’,5’-phosphodiester bonds are hydroit was concluded that cyclic trimer, in which virtually all the lyzed by RNase T2,6v7 while those with 2’,5’-links are not bonds are 3‘,5’-linked, is the main product present in the dimer cleaved. The extent of cleavage by RNase T2 was determined fraction. by HPLC analysis before and after hydrolysis of the RNase T2 The proof of the presence of cyclic trimer in the dimer products with APH. fraction, coupled with the absence of AsppA containing Appreciable yields of unreacted oligomers were detected by oligomers in the trimer to pentamer fractions, led to the the HPLC analysis of the APH hydrolysis products (Table 2). conclusion that cyclic oligomers are present in these fractions These were shown to be cyclic nucleotides by their degradation also. The presence of a mixture of all 3’,5’-linked cyclic to dinucleotides and smaller products on enzymatic hydrolysis tetramer and a cyclic tetramer containing one 2’,5’-linkage was with RNase T2 (Table 3). No adducts of A5’ppA were present deduced from the failure to cleave these isomers with APH and as shown by the absence of AsppA by successive reaction of the formation of AypAp as a reaction product on treatment with the oligomers with RNase T2 and APH. This was an unexpected RNase Tz. A similar approach was used to determine the finding in view of the observation of adducts of A5’ppA in the proportion of all 3’,5‘-linked cyclic pentamer and cyclic penreaction of ImpA in the presence of montmorill~nite.~*~ tamer with only one 2’,5’-linkage in the tetramer fraction (Table Further evidence for the presence of cyclic trimer in the dimer 5). The observation of 35% unreacted material on treatment fraction was obtained by collection of the presumed cyclic trimer of the pentamer fraction with APH led to the conclusion that it from the reverse-phase HPLC column together with ~ A ‘ P A . ~ ~contains 35% cyclic hexamer. Several attempts to resolve these closely eluting compounds The dramatic differences in the structures of the reaction were unsuccessful. A mixture of pA3‘pAand the cyclic trimer products observed when DMAP (3a) and imidazole (ImpA) are was treated with RNase T2,and the hydrosylate was shown to used as activating groups are illustrated in Table 5. Reasons contain pAp, Ap, and A when analyzed by reverse-phase HPLC. for the absence of pyrophosphate groupings in the products Adenosine and a 3% yield of a peak with the same retention formed from 3a and their presence in the oligomers formed from time as A2‘pA (Table 4)were the only products observed after ImpA was investigated. As noted previously, AsppA is formed treating the RNase T2 hydrolysate with APH. From these data in about 6% yield in control reactions of 3a carried out in the (34) Kawamura, K.; Fems, J. P.J. Am. Chem. SOC. 1994, 116, 7564. absence of montmorillonite, but the rate of A5’ppA formation
10920 J. Am. Chem. SOC.,Vol. 116, No. 24, 1994 is slower than oligomer formation. Kinetic analysis of the hydrolysis of 3a under the oligomerization conditions but in the absence of montmorillonite gave a pseudo-first-order rate constant of 2.9 x h-*; a rate equivalent to a half-life of about 24 h (Figure 2). As noted above, the half time for oligomer formation is about 1 h, so hydrolysis in aqueous solution is about 24 times slower than oligomer formation. These data demonstrate that oligomer formation on montmorillonite is complete before appreciable A5‘ppA can form in aqueous solution. The absence of A5’ppA-containingoligomers may also be due to the efficiency with which 3a binds to montmorillonite. About 75% binds within 30 min (Figure 3) while 55% of the ImpA binds in the same time period (Figure 4). The binding and rapid reaction of 3a continues in the subsequent 30 min time interval while the concentration of ImpA in the solution phase stays almost constant. The much more rapid reaction of 3a results in its conversion to oligomers well before the reaction of ImpA is completed. A’ppA forms in the aqueous phase of the ImpA reaction during the relatively slow reaction of ImpA and the dinucleoside pyrophosphate binds to the montmorillonite and is incorporated into the oligomers. The second structural difference observed between the oligomers formed from ImpA and 3a is the high proportion of cyclic nucleotides obtained from 3a (Tables 2 and 5). Cyclic trimer, tetramer, pentamer, and hexamer were observed with an overall yield of 18%. Cyclic dimers and trimers have been reported35 as products from the oligomerization of ImpC and ImpU on U022+,and we observed cyclic trimer from ImpA on montmorillonite (Table 5).’ Formation of cyclic oligomers requires the close proximity of the 5’- and 3’-terminii of the oligomer chain. Molecular modeling studies (CAChe) indicates that an extended form of DMAPpApApA is of lower energy than a conformer in which the 3’- and 5‘-ends are proximate. The solution-phase cyclization of activated linear nucleotides is also considered to be an unlikely possibility because of the absence of a progressive decrease in the yields of cyclic oligomers with increasing nucleotide chain length (Table 2).36 The linear oligomer conformation in which the 3’- and 5’-ends are proximate may be a low-energy conformation on the surface of montmorillonite which results in the formation of cyclic oligomers. The third structural difference between the oligomers formed from 3a and ImpA is the greater regioselectivity for 3’,5’phosphodiester bond formation observed using 3a (Table 5). About 88% of the phosphodiester bonds formed with the (dimethy1amino)pyridinium activating group are 3’3’-linked while about 67% are 3’,5‘-linked when imidazole is the activating group. This regioselectivity of phosphodiester bond formation is greater than the highest previous value of 80% observed in the reaction of 9:l ImpA, A5’ppA mixture^.^ The 2-aminobenzimidazole-activated5‘-AMP(3d) reacted in the presence of montmorillonite to give oligomers containing up to six monomer units. The yield of the oligomers longer (35) Sawai, H.; Higa, K.; Kuroda, K. J. Chem. Soc., Perkin Trans. 1 1992, 505.
(36)Tener, G . M.; Khorana, H. G.; Markham, R.; Pol, E. H. J. Am. Chem. Soc. 1958, 80, 6223.
Prabahar et al. than dimer was very low so it was only possible to characterize the dimer fraction. HPLC analysis on a reverse-phase column separated the mixture into 49% pA2‘pA, 10% pA3’pA,and 41% 3’,5’-cyclic trimer. The presence of cyclic trimer and the absence of A5’ppA-containingoligomers was confirmed by the failure to observe AsppA as a product after the sequential treatment of the dimer fraction with RNase T2 followed by APH. The excess of pA2’pA over pA3’pA, a result that is markedly different from that observed with ImpA, suggests that the higher oligomers are mainly 2’,5’-linked. It is concluded from this study that the best nitrogencontaining leaving groups for nucleotide oligomerization in aqueous solution should have a positive charge or be capable of being protonated on montmorillonite. These positively charged groupings are stabilized by delocalization and solvation. The DMAP group has proven to,be the most effective activating group to date for the formation of longer oligomers which have the highest proportion of 3’3’-linkages.
Conclusions This investigation uncovered a new class of phosphate activating groups, the 4-aminopyridine derivatives (3a-c), which are more effective than imidazole derivatives for the regiospecific formation of 3‘,5‘-linked oligoadenylates on montmorillonite in aqueous solution. Chain lengths as high as dodecamers are formed with a regioselectivity for 3’3’phosphodiester bond formation averaging 88%. The most effective nitrogen-containingactivating groups for the formation of oligonucleotides in aqueous solution contain either a positively charged nitrogen attached to the phosphorus or a nitrogen that is sufficiently basic so that it is protonated by the acidic sites on the montmorillonite. If this positive charge is stabilized by delocalization and solvation, then oligomer formation competes successfully with the hydrolytic cleavage of the activating group and oligomers are formed. The presence of the positive charge is important since this attracts the nucleotide to the negative sites on the montmorillonite resulting in the binding of the activated nucleotide to the catalytic sites there. It should be noted that purines and pyrimidines also contain 4-aminopyridine-type structural units. Some of these compounds, which can be formed from HCN or cyanoacetylene in prebiotic simulation e~periments,3~ may have served as phosphateactivating groups. Acknowledgment. This research was supported by NSF Grant CHE 9301812. HPLC equipment was purchased on NASA Grant NGR-38-018-1148. Timothy D. Cole received summer support as undergraduate research student from NSF and the Howard Hughes Foundation. Alicia Bair who received summer NSF URP support, is thanked for the initial preparation of 3e and 3f, and Dr. John Peyser developed the preparative reverse-phase chromatographic procedures used for the purification of the activated nucleotides. The Unity 500 NMR was purchased with funds from NSF Grant CHE-9105906 and PHS Grant 1 SIORR6245-01. (37) Ferris, J. P.; Hagan, W. J., Jr. Tetrahedron 1984, 40, 1093.
